The variant has been implemented in version 2.0.27. It is not intended as a substitute
of the official rtlinux release from NMT. It can be helpfull and more efficient for those that  
just need a fix ticking but it is less flexible than the original 
version. However it provides some additional services: semaphores, intertask messages and
remote procedure calls that can help in writing complex real time applications.
There is no copyright within the code but I would like to be cited if it used. 

The following comments where provided by Richard Bowser from NMT. They should
be of help both to those using version 2.0.27 and 2.0.29.

Thanks a lot! I got your package and unpacked it into a subdirectory in my
main one. (Since my account is full of differing rtl variants, I don't
dare put one into /usr/src/lilnux unlesss its on a test machine. (A test
machine is defined as one I can crash, which I regularly do quite well.)
I had a few minor problems in getting a running kernel built, but not too
many.

I built my version of your variant on the 0.5a patch to rtlinux-2.0.29, so
I'm not sure how that compares to your kernels. My build blew up on vt.c
in drivers/char because rt_latch2 was undefined (defined in rt_time.c). I
searched the source tree and found the variable otherwise unused, so I
deleted the line assigning it a value. I rebuilt the kernel, but then
things blew up while trying to link in liba: init/main.c was calling
rt_time_init(), which was similarly undefined. Of course, that wasn't
needed any more, so I removed all references to it. After that, things
worked fine. I completed the build successfully. Did you encounter any
problems with unresolved references? Perhaps this was due to my use of the
0.5a patch as opposed to 0.5, which I think you may have used.

I had one more problem after that - your instructions said:

	1 - Apply the rtlinux-0.5 patch but do not do make zImage immediatly;
	2 - Copy the files rt_time.c, rt_time.h and ksyms.c of this
	    distribuition in the appropriate linux directories;
	3 - Carry on the compilation of the kernel by making zImage;
	4 - Copy rt_sched.h of this distribution in the appropiate directory;
	5 - Copy rt_sched.c of this distribution in rt_prio_sched.c in the
	    appropriate directory;
	6 - Do make modules;

I had misinterpreted your #5 above. I did as it said, and copied rt_sched.c 
into rtlinux/kernel. I was rather surprised at first when the make started
to produce a stream of unresolved references. That is, until it occurred
to me that you meant to copy your rt_sched.c over rt_prio_sched.c, thus
causing make to build _your_ scheduler. This was a mistake which came out
of a bad habit of mine, namely being overly literal. However, since you
requested feedback, I felt I could perhaps suggest that mistake could be
prevented if #5 said "Copy rt_sched.c of this distribution into the file
rt_prio_sched.c, thus overwriting it."

Please do not regard this as a complaint: it is not. It's only my suggestion 
for helping "idiot proof" your instructions. (In reality you can't idiot
proof anything because idiots are so *!$@# ingenious.) 

Everything worked well, and I had no real problems along the way. Thanks
again for sending me this material. It will be helpful. I hope what I've
written here will be helpful to you and that the level of detail is not
excessive for you. Thanks again for your courtesy.

Rich B.